Attitude and Heading Reference Systems Questions and Answers

Answer: C

Correct answer: Weights by estimated uncertainty

A Kalman filter carries a covariance matrix representing its current uncertainty in each state, and it weights each new measurement against that uncertainty. The blending is therefore adaptive: when the accelerometers are disturbed by manoeuvring their measurement noise is large and they are trusted less, whereas a fixed complementary filter applies the same crossover regardless.

It also estimates states that are not measured directly, such as the gyro biases themselves. It does require a model of the system - that is what propagates the states between measurements - and it needs neither more sensors nor a higher update rate than the filter it replaces.

Answer: D

Correct answer: A non-zero stationary output

Gyro bias is the output the sensor reports when it is not rotating - a constant, or slowly varying, offset. Because attitude is obtained by integrating rate, a bias integrates into an attitude error that grows linearly with time: a bias of \( 0.01 \) degrees per second gives \( 3 \) degrees after five minutes.

It is distinct from random noise, which integrates to the square-root-of-time growth called angle random walk; from scale factor error, which is proportional to the rate being measured and so appears only during motion; and from response delay, which is a dynamic rather than a static error.

Answer: A

Correct answer: Error growing as the square root of time

Angle random walk describes the attitude error produced by integrating the gyro's white noise. Because independent random increments accumulate as the square root of their number, the resulting angle error grows as \( \sqrt{t} \), and the coefficient is quoted in degrees per root hour.

It is therefore distinguishable from bias, which produces error growing linearly with time, and this difference in growth law is what lets a filter estimate the two separately. A fixed offset is bias, and temperature-induced error is a separate systematic effect usually handled by calibration.

Answer: B

Correct answer: The initial attitude and heading before flight

Alignment establishes the initial attitude and heading from which the system will thereafter integrate. Levelling uses the accelerometers, which sense the gravity vector while the aircraft is stationary, to fix initial pitch and roll; heading comes from the magnetometer or, in a high-grade system, from gyro compassing on the horizontal component of earth rate.

An attitude and heading reference system does not compute position at all, so no position is established. Magnetic variation is looked up from a stored model rather than measured, and air data corrections belong to a separate computer.

Answer: C

Only very low noise gyros can resolve a rate of 15 degrees an hour reliably, which is why the technique belongs to inertial grade systems. It gives true rather than magnetic heading and needs no magnetic sensor at all.

Answer: D

Correct answer: Two axes oscillate at once

Coning arises when the body oscillates about two axes simultaneously with a phase difference between them, so that a body axis traces out a cone. Finite rotations do not commute, and the true attitude change over a cycle is not the integral of the rate components taken one axis at a time - so a naive integration accumulates a systematic error even though the motion returns to where it started.

The remedy is a coning compensation term computed at a high rate, well above the vibration frequencies present. Straight-line acceleration, magnetic disturbance and temperature change all produce errors, but none of them is coning.

Answer: A

The correlation between vibration in rate and in acceleration produces a spurious velocity, so it matters most in a vibrating environment. Like coning it is handled by a high rate compensation algorithm rather than by better sensors.

Answer: B

Correct answer: Small, cheap and rugged

Micro-electromechanical sensors are attractive because they are small, inexpensive and mechanically robust, being etched from silicon in batches. That combination made attitude and heading reference systems affordable in general aviation, unmanned aircraft and consumer devices, where a ring laser or fibre optic gyro would be out of the question on size and cost alone.

The trade is accuracy: they have larger bias, greater bias instability and stronger temperature sensitivity than the optical gyros they displace at the low end. They are therefore not immune to temperature, not free from drift, and not more accurate than a ring laser gyro.

Answer: C

Correct answer: A failed unit can be excluded

Multiple units are fitted for fault tolerance. Two units allow a disagreement to be detected, and three allow the odd one out to be identified by median voting and excluded, so the system continues on the two that agree rather than reverting to no attitude reference at all.

The redundancy is not there to share power or to serve separate displays - a single unit can drive several - and averaging is the wrong operation, since a large error in one unit would drag the average with it. The purpose is to survive a failure, not to improve accuracy.

Answer: D

Correct answer: Flagging when their outputs differ by more than a threshold

Comparison monitoring works by differencing the two units' outputs and raising a warning when the difference exceeds a preset threshold. The threshold is set above the normal spread of two healthy units and below the smallest error that would matter operationally.

With only two units the monitor can say that one of them is wrong but not which, so the usual response is to flag both and require the crew to judge against another source. Internal temperature and supply voltage are built-in test parameters rather than the comparison itself, and the air data computer measures quite different quantities.